Abstract Immune checkpoint inhibitors have markedly improved overall survival in a number of cancers, which has in turn sparked tremendous scientific and financial investment into further expansion of this treatment paradigm. Currently, however, the benefits of immunotherapy have only been realized in a minority of patients. Further complicating the issue, many immunotherapies carry risks of severe adverse immune events, and methods to detect therapeutic efficacy such as anatomical staging and 18F-FDG PET imaging are confounded by the potential presence of immune infiltrate. These invading immune cells can cause potentially responding tumors to increase in size and in 18F-FDG consumption, which make them indistinguishable from progressing malignancies. Because of the lack of current diagnostic capabilities, the only option many patients undergoing immunotherapy have to determine if they are responding is overall survival, which is a long and potentially dangerous approach to determining therapeutic efficacy. Additionally, given the increasing number of drugs and combinations being clinically trialed, the ability to monitor therapeutic efficacy at an earlier stage would potentially help bring new treatments to approval much faster. Currently, there is no approved biomarker for determining therapeutic efficacy, and biopsy analysis of tumor markers such as PD-L1 prior to treatment have only resulted in modest improvements of outcome. Thus a biomarker that predicted response would permit significant advances in both the pre-clinical and clinical investigations. Granzyme B, which is secreted by T effector cells following activation and acts as a potent inducer of apoptosis, is a strong predictor of immunotherapy response. I have developed a novel and selective PET imaging peptide that detects the secreted and active form of granzyme B, permitting differentiation between active response to immunotherapy and non-response in which ?exhausted? T cells that contain granzyme B may be present but are not actively secreting the enzyme. PET imaging with the granzyme B peptide permits highly sensitive and specific prediction of response to immunotherapy prior to changes in tumor volume in murine syngeneic models of cancer. This phenotype is not limited to mice, as human samples analyzed both by antibody and my peptide show significantly higher levels of granzyme B in responding versus non- responding patients. Thus, granzyme B PET imaging offers a unique insight into early response that is not currently possible using any other technique. Current methods cannot accurately define a response prior to destructive sampling, as a response is defined as lack of progression. Given these limitations, I am proposing to use granzyme B PET imaging to stratify mice based on granzyme B levels, followed by biochemical and genetic analysis of responding and non-responding tumors. The non-invasive nature of PET imaging will not only provide accurate differentiation of response prior to any anatomic changes, but will also allow for secondary therapeutic manipulations based on initial PET imaging results.